vulnerability in VMware’s Cloud Director platform that could
potentially allow an attacker to gain access to sensitive
information and control private clouds within an entire
infrastructure.
Tracked as CVE-2020-3956[1], the code injection flaw
stems from an improper input handling that could be abused by an
authenticated attacker to send malicious traffic to Cloud Director,
leading to the execution of arbitrary code.
It’s rated 8.8 out of 10 on the CVSS v.3 vulnerability severity
scale, making it a critical vulnerability.
VMware Cloud Director is a
popular deployment, automation, and management software that’s used
to operate and manage cloud resources, allowing businesses to data
centers distributed across different geographical locations into
virtual data centers.
According to the company, the vulnerability can be exploited
through the HTML5- and Flex-based UIs, the API Explorer interface,
and API access.
The vulnerability impacts VMware Cloud Director versions 10.0.x
before 10.0.0.2, 9.7.0.x before 9.7.0.5, 9.5.0.x before 9.5.0.6,
and 9.1.0.x before 9.1.0.4.
The vulnerability was identified by a Prague-based ethical
hacking firm Citadelo after it was hired earlier this year by an
unnamed Fortune 500 enterprise customer to carry out a security
audit of its cloud infrastructure.
It has also published a proof-of-concept to demonstrate the
exploit’s severity.
“Everything started with just a simple anomaly. When we entered
${7*7} as a hostname for the SMTP server in vCloud Director, we
received the following error message: String value has an
invalid format, value: [49],” Citadelo
noted[4] in its report. “It
indicated some form of Expression Language
injection[5], as we were able to
evaluate simple arithmetic functions on the server-side.”
Using this as an entry point, the researchers said they were
able to access arbitrary Java classes (e.g. “java.io.BufferedReader“) and instantiate
them by passing malicious payloads.
[6]
Citadelo said it was able to perform the following the set of
actions by exploiting the flaw:
- View content of the internal system database, including
password hashes of any customers allocated to this
infrastructure. - Modify the system database to access foreign virtual machines
(VM) assigned to different organizations within Cloud
Director. - Escalate privileges from “Organization Administrator” to
“System Administrator” with access to all cloud accounts by merely
changing the password via an SQL query. - Modify the Cloud Director’s login page, allowing the attacker
to capture passwords of another customer in plaintext, including
System Administrator accounts. - Read other sensitive data related to customers, like full
names, email addresses, or IP addresses.
After Citadelo privately disclosed the findings to VMware on April
1, the company patched the flaws in a series of updates spanning
versions 9.1.0.4, 9.5.0.6, 9.7.0.5, and 10.0.0.2.
VMware has also released a workaround to
mitigate[7] the risk of attacks
exploiting the issue.
“In general, cloud infrastructure is considered relatively safe
because different security layers are being implemented within its
core, such as encryption, isolating of network traffic, or customer
segmentations. However, security vulnerabilities can be found in
any type of application, including the Cloud providers themselves,”
Tomas Zatko, CEO of Citadelo, said.
References
- ^
CVE-2020-3956
(www.vmware.com) - ^
Cloud Director
(www.vmware.com) - ^
proof-of-concept
(github.com) - ^
Citadelo noted
(citadelo.com) - ^
Expression Language injection
(owasp.org) - ^
java.io.BufferedReader
(docs.oracle.com) - ^
workaround to mitigate
(kb.vmware.com)
Read more http://feedproxy.google.com/~r/TheHackersNews/~3/1LPKw-7UuUE/vmware-cloud-director-exploit.html
